Examples Gallery

MurphySig adapts to any context where provenance matters. Here’s how it looks in practice.

Code Comments

JavaScript/TypeScript

/**
 * Signed: Sarah Chen + claude-opus-4-5-20250514, 2026-01-15
 * Format: MurphySig v0.2 (https://murphysig.dev/spec)
 *
 * Context: Implementing OAuth2 token refresh logic. We chose a 5-minute
 * buffer before expiry to account for network latency and clock skew.
 *
 * Confidence: 0.75 - logic is solid, but edge cases around clock drift
 * need production validation
 *
 * Reviews:
 * 2026-03-20 (Sarah + claude-sonnet-4-5): Production hit one clock skew
 * edge case—added 10-minute buffer. New confidence: 0.9
 */
async function refreshTokenIfNeeded(token) {
  // Implementation...
}

Python

"""
Signed: Alex Rivera + gpt-5-20251201, 2026-02-10
Format: MurphySig v0.2 (https://murphysig.dev/spec)

Context: ML pipeline for user intent classification. Training on 50k
samples, using DistilBERT for speed over accuracy (product decision).

Confidence: 0.6 - accuracy is 87% but we haven't validated on edge cases
like multilingual input or highly ambiguous queries

Heuristic: We're prioritizing latency (p95 < 100ms) over accuracy here.
If accuracy drops below 85% in production, switch to full BERT.
"""
class IntentClassifier:
    # Implementation...

Rust

/// Signed: Jordan Lee (solo), 2026-01-20
/// Format: MurphySig v0.2 (https://murphysig.dev/spec)
///
/// Context: Lock-free concurrent queue implementation. Heavily inspired
/// by Dmitry Vyukov's MPMC queue but adapted for our use case (single
/// producer, multiple consumers with priority).
///
/// Confidence: 0.4 - compiles, passes basic tests, but concurrency bugs
/// are sneaky. Needs fuzzing and formal verification before production.
///
/// Reflections:
/// 2026-06-15: After 3 months in production with zero issues, I now
/// understand why Vyukov's design is so elegant. Confidence would be 0.9
/// if I wrote this again.
pub struct SPMCQueue<T> {
    // Implementation...
}

Document Headers

Markdown Technical Spec

# Authentication System Architecture

*Signed: Platform Team (Maria, Carlos, Jamie) + claude-opus-4-5-20251101, 2026-01-08*
*Format: MurphySig v0.2*

*Context: Post-security audit redesign. Moving from JWT to session tokens
with Redis backing. Decision made after weighing XSS risks vs. scalability.*

*Confidence: 0.8 - architecture is sound, but migration path for existing
users needs more detail*

## Overview
...

Research Paper

% Signed: Dr. Emma Torres + claude-opus-4-5-20250514, 2025-12-01
% Format: MurphySig v0.2 (https://murphysig.dev/spec)
%
% Context: AI-assisted literature review and analysis. I wrote the core
% argument and interpretations; Claude helped synthesize 80+ papers and
% catch inconsistencies in citations.
%
% Confidence: 0.85 - argument is novel and well-supported, but the neural
% architecture comparison in Section 4.2 may oversimplify recent advances

\documentclass{article}
\title{Emergent Reasoning in Transformer Architectures}
...

Git Commit Messages

Feature Commit

feat: add real-time collaboration with WebSockets

Signed: Dev Team + cursor-0.42, 2026-01-12
Format: MurphySig v0.2

Context: Users requested Google Docs-style live editing. Chose
Socket.IO over raw WebSockets for fallback support (some corporate
firewalls block WS). Implemented operational transforms for conflict
resolution.

Confidence: 0.65 - happy path works well, but conflict resolution
under high concurrency needs load testing

Files changed: 12
Lines: +847, -23

Bug Fix Commit

fix: resolve race condition in payment processing

Signed: Maya Patel + claude-sonnet-4-5-20251015, 2026-02-18
Format: MurphySig v0.2

Context: CRITICAL BUG - users reported duplicate charges. Root cause
was insufficient transaction isolation. Added database-level locks
and idempotency keys.

Confidence: 0.95 - fix is straightforward, added integration tests,
verified in staging with production-scale load

Reviews:
2026-02-20 (Maya): 48 hours in prod, zero duplicate charges, perf
impact negligible. Confidence confirmed at 0.95.

Creative Work

Blog Post

---
title: "Building in Public: Lessons from 6 Months of Shipping Daily"
author: Kev Murphy
date: 2026-01-10
---

*Signed: Kev Murphy + claude-opus-4-5-20251101, 2026-01-10*
*Format: MurphySig v0.2*

*Context: Reflecting on my build-in-public journey. Claude helped
structure my scattered notes and tighten the prose, but the stories
and lessons are mine.*

*Confidence: 0.9 - I stand behind this. The writing feels honest.*

*Reflections:*
*2026-06-10: Re-reading this 5 months later—I was more burned out than*
*I admitted here. The optimism was real, but incomplete.*

---

Six months ago, I committed to shipping something every day...

Design System Documentation

# Button Component API

*Signed: Design System Team + cursor-0.44 + claude-opus-4-5, 2026-01-05*
*Format: MurphySig v0.2*

*Context: Multi-model collaboration. Cursor helped scaffold the component
structure, Claude reviewed for accessibility compliance. Final decisions
on variant naming and API design were team consensus.*

*Confidence: 0.7 - API feels clean, but we haven't validated with product
teams yet. Expecting iteration based on real-world usage.*

## Usage

...

Standalone Files

Configuration File

# deployment/staging.yaml

# Signed: DevOps + gpt-5-20251201, 2026-01-22
# Format: MurphySig v0.2
#
# Context: Staging environment config for Kubernetes. Auto-generated
# from production config with scaling adjustments (2 replicas instead
# of 10, smaller resource limits).
#
# Confidence: 0.8 - tested in dev, should work in staging, but cloud
# provider quota limits might need adjustment

apiVersion: apps/v1
kind: Deployment
...

Dataset Metadata

{
  "dataset": "user-feedback-2026-q1",
  "signed": "Data Team (Lin, Marcus) + claude-opus-4-5-20251101",
  "date": "2026-03-31",
  "format": "MurphySig v0.2",
  "context": "Quarterly user feedback analysis. Manually labeled 5k responses, Claude helped identify themes and edge cases in remaining 45k. Human review on 10% sample showed 94% agreement.",
  "confidence": 0.85,
  "confidence_text": "High agreement on labeling, but 'neutral' category had most ambiguity",
  "rows": 50000,
  "columns": ["user_id", "feedback_text", "sentiment", "theme", "priority"]
}

Multi-Model Evolution

/**
 * Signed: Jamie + cursor-0.41 + claude-opus-4-5-20251101, 2026-01-10
 * Format: MurphySig v0.2
 *
 * Context: Search autocomplete feature. Cursor scaffolded the debounce
 * logic, Claude reviewed for edge cases (empty queries, special chars).
 * I wrote the API integration and caching strategy.
 *
 * Confidence: 0.7 - works well for English queries, untested on CJK input
 *
 * Reviews:
 * 2026-01-15 (Jamie + gpt-5): Added CJK tokenization. Tested with Japanese
 * and Chinese queries—works but needs better ranking. Confidence: 0.75
 *
 * 2026-02-01 (Jamie + claude-sonnet-4-5): Rewrote ranking algorithm based
 * on user analytics. Click-through rate improved 23%. Confidence: 0.9
 */
class SearchAutocomplete {
  // Implementation...
}

Text Confidence (No Numerical Score)

"""
Signed: Research Team + claude-opus-4-5-20250514, 2025-12-15
Format: MurphySig v0.2

Context: Exploratory analysis of A/B test results. Early-stage investigation,
not production code.

Confidence: Low - statistical significance is marginal (p=0.048), sample size
was smaller than planned due to early termination, and we haven't controlled
for confounding variables like time-of-day effects

Note: Use this analysis to inform next experiment, not for product decisions
"""
def analyze_experiment_results(data):
    # Implementation...

Key Patterns

Across these examples, notice:

Natural language over technical jargon — Anyone can read it
Confidence calibration — Low confidence is valuable information
Context explains “why” — Not just what changed, but why it was built this way
Reviews show evolution — The work gets better over time
Reflections add humanity — The maker’s voice, not just the code

MurphySig adapts to your workflow. Use what fits. Skip what doesn’t. The goal is honest provenance, not bureaucracy.

Ready to start? Read the Specification →

Signed: Kev Murphy + claude-sonnet-4-5-20260106, 2026-01-06 Format: MurphySig v0.2

Context: Examples gallery to showcase MurphySig across different contexts and languages. Demonstrates flexibility and real-world applicability.

Confidence: 0.8 - examples are realistic and diverse, but could use more edge cases (e.g., scientific notebooks, legal documents)